Saved in:
Bibliographic Details
Main Authors: Lee, JunKyu, Varghese, Blesson, Vandierendonck, Hans
Format: Preprint
Published: 2022
Subjects:
Online Access:https://arxiv.org/abs/2210.16083
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910426799800320
author Lee, JunKyu
Varghese, Blesson
Vandierendonck, Hans
author_facet Lee, JunKyu
Varghese, Blesson
Vandierendonck, Hans
contents This paper analyzes the effects of dynamically varying video contents and detection latency on the real-time detection accuracy of a detector and proposes a new run-time accuracy variation model, ROMA, based on the findings from the analysis. ROMA is designed to select an optimal detector out of a set of detectors in real time without label information to maximize real-time object detection accuracy. ROMA utilizing four YOLOv4 detectors on an NVIDIA Jetson Nano shows real-time accuracy improvements by 4 to 37% for a scenario of dynamically varying video contents and detection latency consisting of MOT17Det and MOT20Det datasets, compared to individual YOLOv4 detectors and two state-of-the-art runtime techniques.
format Preprint
id arxiv_https___arxiv_org_abs_2210_16083
institution arXiv
publishDate 2022
record_format arxiv
spellingShingle ROMA: Run-Time Object Detection To Maximize Real-Time Accuracy
Lee, JunKyu
Varghese, Blesson
Vandierendonck, Hans
Computer Vision and Pattern Recognition
This paper analyzes the effects of dynamically varying video contents and detection latency on the real-time detection accuracy of a detector and proposes a new run-time accuracy variation model, ROMA, based on the findings from the analysis. ROMA is designed to select an optimal detector out of a set of detectors in real time without label information to maximize real-time object detection accuracy. ROMA utilizing four YOLOv4 detectors on an NVIDIA Jetson Nano shows real-time accuracy improvements by 4 to 37% for a scenario of dynamically varying video contents and detection latency consisting of MOT17Det and MOT20Det datasets, compared to individual YOLOv4 detectors and two state-of-the-art runtime techniques.
title ROMA: Run-Time Object Detection To Maximize Real-Time Accuracy
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2210.16083